Student Research: Julia M. Gohlke

PhD, , 2004
Faculty Advisor: Elaine M. Faustman

A Quantitative Examination of Ethanol-Induced Neurodevelopmental Toxicity Using Computational Models


Abstract

Computational, systems-based approaches can provide a quantitative construct for evaluating risk in the context of mechanistic data. Computational models are developed in the rat, mouse, rhesus monkey, and human, describing the acquisition of adult neuron number in the neocortex during the key neurodevelopmental processes of neurogenesis and synaptogenesis. Models are based on experimentally derived parameters including division, transformation, and death rates and are critically compared to independent, stereologically determined neuron number data in the adult rat, mouse, rhesus monkey and human. Mechanistic data from the rat describing ethanol-induced toxicity in the developing neocortex is applied to these models. Our model can explain long-term neocortical neuronal loss after in utero exposure to ethanol based on inhibition of proliferation during neogenesis. Modeling ethanol-induced apoptosis during synaptogenesis predicts similar neuronal loss, but at much higher daily peak blood ethanol concentrations (BEC), 500 mg/dl (10-15 drinks) compared with 150 mg/dl (3-5 drinks) in the neurogenesis model. Unlike effects seen after exposure during neurogenesis, ethanol-induced apoptosis during synaptogenesis does not correlate with lasting effects on neocortical neuron number as indicated by stereological data in the adult rat. Our models suggest primate species may be more sensitive to ethanol-induced effects on neurogenesis based on the increased duration of neurogenesis in primates. Our model predicts a 30% neuronal deficit after daily peak BECs reaching 20 mg/dl, or approximately 1 drink/day in the human, whereas peak daily BECs of 100 mg/dl are necessary to predict similar deficits in the rat. The computational model presented here also provides a framework for evaluating both experimental uncertainty and biological variability. Variability in parameters is estimated by setting output variability equal to the reported variability in stereological estimates of adult neocortical neuron number. Model simulations suggest true biological variation in proliferation, transformation and death rates is low (CV ≤ 2.5%). However, variation reported in experimental studies for parameter estimation is large (CV 10-30%), suggesting ascertainment of intra-individual correlations in rates over time is necessary to understand the contributions of biological variability versus experimental uncertainty. In a broader context, this research serves as a robust, quantitative evaluation of mechanistic hypotheses for neurodevelopmental toxicity.